基于改进的压缩式随机游走算法对静态电源/地网的模拟
Simulation of Static Power/Ground Networks with Improved Compressed Random Walk Algorithm
-
摘要: 将随机游走法和等效电路压缩相结合,对静态P/G网(Power and Ground Networks)进行分析.针对一个大规模的电路,在经过多层的参数提取和建模后,得到静态P/G网模型.首先根据网络的规律性,运用等效电路压缩法将原始的P/G网进行压缩处理,然后运用随机游走法求解,最后利用计算得到的化简网络电压值,通过相关的插值公式得到原网络的电压值.实验数据表明,改进的压缩式随机游走法可有效简化网络的复杂性,节省计算时间,计算速度提高到普通随机游走法的两个数量级以上.Abstract: The static power and ground(P/G) network is analyzed by a combination of compressed equivalent circuit modeling and random wall algorithm.A model of power and ground network is obtained by parameter extraction and modeling over whole chips.The method builds an equivalent model for original network and uses random walk method to solve the simplified network.As a result,the improved compressed random walk algorithm saves CPU time greatly.The speed of the algorithm is more than two order of magnitude faster than the normal random walk algorithm.